• Joined on 2025-11-19
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 21:46:56 -05:00
2d6fe2c962 autoresearch: phase1 trial 5 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 19:56:34 -05:00
c8a495dd22 fix: reward v4 — full sim bypass kills circular driving at root
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 18:33:09 -05:00
7b8830f0cb milestone: Phase 1 complete — genuine driving confirmed; launch Phase 2 corner learning
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 18:18:01 -05:00
cb82121e98 autoresearch: phase1 trial 50 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 18:17:58 -05:00
3cbe4bd26e autoresearch: phase1 trial 50 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 17:15:33 -05:00
4c9b68dd47 autoresearch: phase1 trial 40 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 16:28:21 -05:00
ed65cf5997 autoresearch: phase1 trial 30 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 15:38:19 -05:00
29a45e017b autoresearch: phase1 trial 20 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 15:00:27 -05:00
caf91c9fe6 autoresearch: phase1 trial 10 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 14:28:09 -05:00
87cff0c9b7 autoresearch: phase1 trial 40 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 14:13:23 -05:00
1734e1359e autoresearch: phase1 trial 30 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 13:41:57 -05:00
362c616457 autoresearch: phase1 trial 20 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 13:08:01 -05:00
cdb7b80494 autoresearch: phase1 trial 10 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 12:36:19 -05:00
fcb6ea1ac2 fix: path-efficiency reward (v3) defeats circular driving exploit
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 12:11:07 -05:00
d25bc71008 autoresearch: phase1 trial 10 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 11:27:51 -05:00
5e93dae316 fix: hack-proof reward shaping + reward hacking detection + research log
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 11:01:19 -05:00
0c6263352b autoresearch: phase1 trial 10 results
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 10:17:11 -05:00
8c9fd76c68 fix: reduce timesteps to 1k-5k for Phase 1 CPU training; add sim health/stuck detection; fix PPO throttle clamp
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 09:03:18 -05:00
c804189dd0 feat: Wave 1 complete — real PPO training, model save, GP+UCB autoresearch, 37 tests passing
paulh pushed to master at paulh/donkeycar-rl-autoresearch 2026-04-13 00:56:09 -05:00
083326a497 AUTORESEARCH: 300 total trials complete - best mean_reward=141.85 at n_steer=8, n_throttle=5, lr=0.00202